ts-web-scraper/dist/pagination.d.ts

A powerful web scraper for both static and client-side rendered sites using only Bun native APIs

/**
 * Detect pagination type and extract info from HTML
 */
export declare function detectPagination(html: string, currentUrl: string): PaginationInfo;
/**
 * Extract page number from URL
 */
export declare function extractPageFromUrl(url: URL): number | null;
/**
 * Generate URL for a specific page
 */
export declare function generatePageUrl(baseUrl: string, page: number, pattern?: PaginationPattern): string;
/**
 * Extract all page URLs from pagination
 */
export declare function extractAllPageUrls(paginationInfo: PaginationInfo): string[];
/**
 * Check if pagination has more pages
 */
export declare function hasMorePages(paginationInfo: PaginationInfo): boolean;
/**
 * Get next page URL
 */
export declare function getNextPageUrl(paginationInfo: PaginationInfo, baseUrl?: string): string | null;
/**
 * Auto-paginate and collect all pages
 */
export declare function autoPaginate(startUrl: string, fetcher: (url: string) => Promise<{ html: string, url: string }>, options?: {
    maxPages?: number
    delay?: number
  }): AsyncGenerator<{ html: string, url: string, pageNumber: number }>;
export declare interface PaginationInfo {
  type: PaginationType
  currentPage?: number
  totalPages?: number
  nextUrl?: string
  prevUrl?: string
  pageUrls?: string[]
  hasMore?: boolean
  cursorNext?: string
  cursorPrev?: string
}
export declare interface PaginationPattern {
  selector?: string
  urlPattern?: RegExp
  pageParam?: string
}
/**
 * Pagination Auto-Detection
 *
 * Detect and handle various pagination patterns
 * Uses ONLY Bun native APIs - no external dependencies!
 */
export declare enum PaginationType {
  NONE = 'none',
  PAGE_NUMBERS = 'page_numbers',
  NEXT_PREV = 'next_prev',
  LOAD_MORE = 'load_more',
  INFINITE_SCROLL = 'infinite_scroll',
  CURSOR = 'cursor',
}